Lessons learned from attending the Microsoft AI Strategy Bootcamp

By costansin
Microsoft Azure AI and Digital Health

This week we were invited for a technical and business workshop organized by Microsoft AI and Emerging Tech Team. This group was on a mission in Malta to meet up with various Government and private entities and evangelize Microsoft’s Artificial Intelligence Roadmap.

This article documents the current AI climate and what tools are being made available by Microsoft to start doing something.

Azure Databricks ecosystem

Bootcamp in a Nutshell

Day 1: Technical Hands-on Workshop

Given my background in Cloud Computing, Open Source Tech, Data, and AI, this was perhaps my favorite day of the Bootcamp. We went head-first using Microsoft Azure Cognitive Services and explored how we can use text, speech, image, knowledge, search among other skillsets, to use AI to solve business problems.

We went on to also try the Azure Databricks service, which sounded really familiar and similar to AWS EMR service. They both use Apache Spark (well, to be precise, AWS offers Apache Flink as well) to run Big Data stream and batch processing. Despite the tech disadvantage of Databricks, I found that Azure as a whole, was more focused on giving us the whole AI-as-a-Service experience. With AWS, you are merely given a bucnh of Data Science components, and then it’s up to you to figure out the best strategy.

When discovering all these Azure services in the AI toolkit chain, I also noticed how Azure Big Data services such as HDInsight and CosmosDB were based on existent open-source technologies. HDInsight is basically Apache Hadoop and Kafka, while CosmosDB felt strangely familiar with MongoDB..

LESSONS LEARNED:

  1. We managed to successfully use Image Recognition and other Cognitive Services to build a basic Machine Learning model and successfully predicting events using the Databricks AI service.
  2. Implementing a working AI model took us just under 2 hours, including setting up the environment from scratch. And this is probably where the whole shebang of Azure strategy is. Democratizing AI, should perhaps be the title of Microsoft’s AI Strategy presentation (for next time folks!) as they are really making a good job in making AI available to everyone at an affordable on-demand pricing model.

Day 2: AI Applicability in Society

The second day was more geared towards people with no background in AI or machine learning and the Microsoft Team delivered a presentation based on AI applicability.

However, it got interesting again when they started talking about how AI and Digital Health has improved, especially given the massive Cloud computing resources that are available. Microsoft Azure did a very good job in penetrating this field, and embellishing their AI toolkit with Digital Health in mind.

Here are some ideas of how AI can be applied at a national level:

  • Anomaly detection for Tax or Insurance Fraud
  • Citizen sentiment analysis to gauge efficacy of new national policies
  • Faster and more efficient Health Services
  • Faster and more efficient Customer Care and Ministerial enquiry processes

LESSONS LEARNED:

  1. Most countries are finally accepting that AI is real and useful, and not science fiction or hype.
  2. AI can provide value equally for Governments as it does for commercial industry.
  3. Microsoft AI has been positioning itself with Digital Health in mind.

AI Challenges in Malta

Like in most workshops, we always feel inspired, believing in dragons and unicorns once again. However this should not discount our awareness of the challenges in the current AI climate in Malta.

The great news is that an AI strategy taskforce is already setup in Malta, and a high-level AI Policy is in currently progress. However, like most European countries, we are still lagging in adopting AI with society. Data Governance, AI Ethical frameworks, are among the AI challenges which take time and cultural changes to take place, despite the technology has matured.

Despite the challenges, however, commercial entities should not just wait for the acceptance cycle to kick in. This is the right time to innovate, learn, and exploring AI seriously.

If you need assistance in exploring AI or similar emerging technologies, by building demos or minimum viable products (MVPs) that you can show to your investors, do get in touch with us.